A multi-cell UWB-IR localization system for robot navigation

UWB-IR is an excellent choice for high precision indoor localization. However due to its short range, a multi- cell based implementation is necessary to cover large indoor area. Such multi-cell systems require the use of robust techniques to overcome issues that adversely affect the localization accuracy at inter-cell boundaries. In this paper, the techniques used and experimental results obtained for a multi-cell UWB-IR robot navigation system are discussed. A robust methodology based on boundary detection and path prediction is proposed. Using the proposed methodology, an RMS localization estimation error of less than 15 cm is achieved at all positions within the localization space including the inter-cell boundaries.